A Comparison of Utilized and Theoretical Covariance Weighting Matrices on the Estimation Performance of Quadratic Inference Functions

The quadratic inference function (QIF) method is increasingly popular for the marginal analysis of correlated data due to its advantages over generalized estimating equations. Asymptotic theory is used to derive analytical results from the QIF, and we, therefore, study three asymptotically equivalent weighting matrices in terms of finite-sample parameter estimation. Furthermore, to improve small-sample estimation, we study modifications to the estimation procedure. Examples are presented via simulations and application. Results show that although theoretical weighting matrices work best, the proposed estimation procedure, in which initial estimates are held constant within the matrix of estimated empirical covariances, is preferable in practice.

[1]  Yang Bai,et al.  Penalized quadratic inference functions for single-index models with longitudinal data , 2009, J. Multivar. Anal..

[2]  A. U.S.,et al.  Generalized method of moments estimation for linear regression with clustered failure time data , 2009 .

[3]  Incorporating Correlation for Multivariate Failure Time Data When Cluster Size Is Large , 2010, Biometrics.

[4]  Min Zhu,et al.  Efficient parameter estimation in longitudinal data analysis using a hybrid GEE method. , 2009, Biostatistics.

[5]  The effect of cluster size imbalance and covariates on the estimation performance of quadratic inference functions , 2012, Statistics in medicine.

[6]  R Z Omar,et al.  Bayesian methods of analysis for cluster randomized trials with binary outcome data. , 2001, Statistics in medicine.

[7]  Catherine Loader,et al.  Iteratively Reweighted Generalized Least Squares for Estimation and Testing With Correlated Data: An Inference Function Framework , 2007 .

[8]  Christopher G. Small,et al.  Hilbert Space Methods in Probability and Statistical Inference: Small/Hilbert , 1994 .

[9]  B. Lindsay.  Conditional score functions: Some optimality results , 1982 .

[10]  Peter X-K Song,et al.  Quadratic inference functions in marginal models for longitudinal data , 2009, Statistics in medicine.

[11]  P. McCullagh,et al.  Generalized Linear Models , 1972, Predictive Analytics.

[12]  F. Windmeijer.  A Finite Sample Correction for the Variance of Linear Two-Step GMM Estimators , 2000 .

[13]  L. Hansen,et al.  Finite Sample Properties of Some Alternative Gmm Estimators , 2015 .

[14]  Eric R. Ziegel,et al.  Generalized Linear Models , 2002, Technometrics.

[15]  B. Lindsay,et al.  Improving generalised estimating equations using quadratic inference functions , 2000 .

[16]  B. Leroux,et al.  Analysis of clustered data: A combined estimating equations approach , 2002 .

[17]  S. Zeger,et al.  Longitudinal data analysis using generalized linear models , 1986 .

[18]  M. Puumala,et al.  Optimal combination of estimating equations in the analysis of multilevel nested correlated data , 2010, Statistics in medicine.

[19]  P. X. Song,et al.  A note on improving quadratic inference functions using a linear shrinkage approach , 2011 .

[20]  Runze Li,et al.  Quadratic Inference Functions for Varying‐Coefficient Models with Longitudinal Data , 2006, Biometrics.

[21]  L. Hansen.  Large Sample Properties of Generalized Method of Moments Estimators , 1982 .

[22]  J. Robson,et al.  Improving uptake of breast screening in multiethnic populations: a randomised controlled trial using practice reception staff to contact non-attenders , 1997, BMJ.

[23]  Dylan S. Small,et al.  Marginal regression analysis of longitudinal data with time‐dependent covariates: a generalized method‐of‐moments approach , 2007 .

[24]  A bias‐corrected covariance estimate for improved inference with quadratic inference functions , 2012, Statistics in medicine.

[25]  Zhongyi Zhu,et al.  Partial Linear Models for Longitudinal Data Based on Quadratic Inference Functions , 2008 .

[26]  C. Small,et al.  Hilbert Space Methods in Probability and Statistical Inference , 1994 .